Remaining useful life (RUL) is an essential part of prognostic and health management, which can be employed to enhance the reliability and reduce the maintenance cost. For many products, due to the change of external operating conditions and internal mechanisms, their degradation trajectories tend to exhibit two-phase patterns. In the current method, linear Wiener process is used in each phase for degradation modeling and RUL prediction. In practice, the degradation process of each phase exhibits nonlinear characteristics, where using linear Wiener process to establish two-phase model is often inadequate. In this paper, a novel approach for two-phase degrading product is proposed. A degradation model using nonlinear Wiener process is adopted to characterize the two-phase degradation trajectory firstly. The maximum likelihood estimation (MLE) is used to estimate the unknown parameters of proposed model and Bayesian method is employed for updating the parameters. Taking into account the randomness of the initial state transition to the changing state and the variability in different units degradation, the approximate analytical solution of RUL under the concept of the first passage time is derived. Finally, the effectiveness of the proposed RUL prediction method is demonstrated through a simulation study and a turbofan engine degradation dataset.

剩余使用寿命（RUL）是预测和健康管理的重要组成部分，可用于提高可靠性和降低维护成本。对于许多产品而言，由于外部操作条件和内部机制的变化，其降解轨迹往往呈现出两阶段模式。在当前的方法中，线性维纳过程用于每个阶段的退化建模和 RUL 预测。在实际应用中，每一相的退化过程都表现出非线性特征，使用线性维纳过程建立两相模型往往是不够的。在本文中，提出了一种用于两相降解产物的新方法。首先采用非线性Wiener过程的退化模型表征两相退化轨迹。最大似然估计（MLE）用于估计所提出模型的未知参数，并采用贝叶斯方法更新参数。考虑到初始状态向变化状态转变的随机性和不同单元退化的可变性，推导出首次通过时间概念下RUL的近似解析解。最后，通过仿真研究和涡扇发动机退化数据集证明了所提出的 RUL 预测方法的有效性。推导出第一次通过时间概念下RUL的近似解析解。最后，通过仿真研究和涡扇发动机退化数据集证明了所提出的 RUL 预测方法的有效性。推导出第一次通过时间概念下RUL的近似解析解。最后，通过仿真研究和涡扇发动机退化数据集证明了所提出的 RUL 预测方法的有效性。